-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Ryzen 5000 series overclocking
- Thread starter Guest2
- Start date
I think its more likely the new 8+8 layout that's allowed the infinity fabric to go higher. 4+4+4+4 means more transfers across the fabric bus which I assume/guess is "worse" for high bus speeds. Whether that counts as an architecture change or not I have no idea, I'm just interested in the possibilities for performance increasing
.Without per core overclocking then an 8+8 layout could actually be a limiting factor as now a cluster of 8 cores would be limited to the speed/voltage requirements of the lowest quality core instead of a cluster of 4 cores.
1usmus seems to think so and is apparently already coding CTR for it.
https://twitter.com/1usmus/status/1311988058825293825
Associate
- Joined
- 19 Jan 2007
- Posts
- 1,813
- Location
- Up a Tree
That sounds very promising.
exciting and very time consuming.
I'll probably just set what I think is a safe voltage like 1.3 and then find the stable all core clock and from there start tweaking each core to find the maximum clocks. Followed by ram tweaking - it's going to take a week just to overclock the system isn't it, sigh
I think its more likely the new 8+8 layout that's allowed the infinity fabric to go higher. 4+4+4+4 means more transfers across the fabric bus which I assume/guess is "worse" for high bus speeds. Whether that counts as an architecture change or not I have no idea, I'm just interested in the possibilities for performance increasing
Without per core overclocking then an 8+8 layout could actually be a limiting factor as now a cluster of 8 cores would be limited to the speed/voltage requirements of the lowest quality core instead of a cluster of 4 cores.
IF increase may well be fake news, I now know of 2 5600x that can't go over 1900fclk. See Buildzoid's latest YT vid (Actually Hardcore Overclocking channel for those who haven't seen his stuff before) for the latest one. Its possible this is down to an AGESA update being needed, but it's not looking promising at all. This changes my plans significantly, if there's literally no difference in memory oc between 3000 and 5000 I may as well go small on Ryzen 5000 for the IPC uplift and go big when AM5/DDR5 drops. Glad I kept the new CH8 Impact in its box anyway, its just an expensive sidegrade if I'm not getting a 16 core.
All this secrecy over new releases is wearing very thin now, I miss the days when hardware reviews came out up to a month before the hardware was publicly available. The only people who benefit from review embargos until the hour of release are scalpers and trolls.
The initial anecdotal evidence that the 5600x IF / Memory controller isn't clocking well past 1900 (buildzoid & jumper118) somewhat contradicts the leaked slide from AMD about clocking it to 2000. I say somewhat because they of course had the huge caveat of "good luck" e.g. the silicon lottery.
So that begs the question, are the GloFo fabbed IO chips going to be binned? And if so, are they more likely to assemble better binned parts onto the more expensive SKU's?
So that begs the question, are the GloFo fabbed IO chips going to be binned? And if so, are they more likely to assemble better binned parts onto the more expensive SKU's?